The existence of tumor-initiating cancer stem cells (CSC) has been shown in a variety of solid tumors (e.g., breast, prostate, glioblastoma, liver). However, these CSC have highly variable antigenic and functional properties even when derived from the same tumor. These observations highlight a cardinal problem in CSC biology, namely, the heterogeneity of the CSC. Different mechanisms could explain the origin and heterogeneity of CSC such as (i) differentiation arrest (stem cells), (ii) dedifferentiation (mature cells) and (iii) transdifferentiation (bone marrow stem cells). It is conceivable that all 3 mechanisms may be corrupted by oncogenic events, resulting in an assortment of CSC and explaining their heterogeneity. Defining and characterizing this heterogeneity is of vital importance for understanding CSC biology, and for effective therapeutic translation. Although a variety of different stem cell markers have been used to isolate and characterize CSC from hepatocellular carcinoma (HCC), none seem to be specific for liver malignancies and none of the isolated fractions show uniform properties. Therefore we opted to use a functional approach to isolate CSC based on their ability to efflux Hoechst dye via ABC transporters. This population is referred to as side population (SP). We consider the functional approach as less biased since it identifies the stem cells by their unique property of chemo resistance. Given that epigenetic regulation plays a crucial role both in stem cell and cancer development, we chose epigenic modulation as an additional tool to narrow down the heterogeneity of SP-fraction. The specific objectives of this study are to: (i) characterize liver CSC heterogeneity using in vitro and in vivo functional assays;and (ii) investigate the effect of epigenetic modulation on the CSC by treatment with the DMNT1-inhibitor Zebularine (ZEB). Our recent results include: (i) Epigenetics are critical in stem cell biology, promoting both pluripotency of stem cells and differentiation of more mature derivatives. If similar mechanisms are relevant for the CSC model, then epigenetic modulation might enrich the CSC population and facilitate CSC isolation and rigorous evaluation. We have evidence that treatment of primary human liver cancer cell lines and established cancer cells with ZEB increased the frequency of cells with CSC properties within the SP as judged by self- renewal, superior tumor-initiating capacity in serial transplantations and direct cell tracking experiments. Further, integrative transcriptome analysis of ZEB-enriched CSC populations show common traits enriched for stemness-associated gene sets, although each individual gene expression signature exhibited activation of different oncogenic pathways (e.g., EGFR, SRC and MYC). The common CSC signature was associated with poorly differentiated tumors, high recurrence and poor survival in liver and other types of cancers;and (ii) we have also exploited the epigenetic modulation of the cancer epigenome by Zeb in conjunction with the ability of stem/progenitor cells to grow as non-adherent spheres in serum-free media to provide a further enrichment of the CSC fraction. HCC cell lines were plated in high- and low-density (HD and LD, respectively) conditions and treated with 100 M Zeb for 3 days. The sphere-forming ability and serial sphere formation were investigated in Matrigel and suspension culture, respectively. Genes that are preferentially expressed in cells grown under sphere conditions were examined by qRT-PCR and surface markers by FACS. The HCC cell line Huh7 grown under LD failed to form tumor spheres when placed in Matrigel or non-adherent growth conditions. In striking contrast, Zeb-treated LD cells were capable of forming spherical clonal expanding colonies that showed stem cell-like properties such as strong induction of the stem cell-related genes Nanog and OCT4, and did not express EpCam. Cells grown at HD possessed a similar sphere-forming ability regardless of Zeb pretreatment. However, morphological characterization indicated a higher variability in sphere size with a predominance of smaller sized spheres, suggesting reduced proliferation potential and/or differences in cell type of origin, with smaller spheres originating from the progenitors, and larger spheres from the stem cells. Consistent with this, HD spheres showed a relatively smaller induction of Nanog and OCT4, expressed high levels of EpCam and exhibited reduced self-renewal in a serial sphere formation assay. The sphere-forming ability of LD and HD cells was associated with the presence of a subpopulation of CD133-/cKit+ cells. These results suggest that the T double selective approach offers an in vitro model to enrich for cells with CSC properties. Both cell density and epigenetic modulation were able to affect the sphere-forming potential of Huh7 cells. Further studies are in progress to characterize the HD and LD cell population in more detail and to clarify their tumorigenic potential and capacity to propagate tumor growth in vivo. Future plans include (i) characterizing the antigenic properties of the CSC before and after ZEB treatment;(ii) defining specific gene expression signatures using microarray analysis;(iii) integrating gene expression signatures with copy number variation (CNV) analysis obtained from the same groups;(iv) characterizing the effect of epigenetic modulation on the CSC by methylation-specific pyrosequencing;(v) using primary human liver tumors to validate our ongoing work in the HCC cell lines;(vi) testing the usefulness of these integrative specific genomic signatures for prognostic prediction, both retrospectively and prospectively, in cohorts of HCC patients available to us at the NCI;and (vii) using the integrative genomic signatures to identify novel and specific molecular therapeutic targets for the HCC-derived CSC. We are also addressing the mechanisms of cancer development in the liver. It is generally accepted that any cell that can proliferate repeatedly--and this includes all cells within the lineage from hepatoblast and adult liver stem cell to hepatocyte and cholangiocyte--is susceptible to neoplastic transformation from a variety of carcinogenic insults. In addition to mature hepatocytes and cholangiocytes, incompletely differentiated progeny of adult liver stem cells can give rise to primary liver cancer (PLC). These considerations led us to hypothesize that the PLC phenotypes are dictated primarily by the differentiation stage at which transformation occurs and comprise a spectrum of neoplasms, ranging from well-differentiated HCC to well-differentiated intrahepatic cholangiocellular carcinoma (ICC). The concept that HCC and ICC are closely related is not new since hepatic neoplasms composed of the cells expressing morphological characteristics of both HCC and ICC were first described more than 100 years ago. To test this hypothesis we plan to use an approach based on in vitro transformation followed by in vivo transplantation (using murine xenograft models) of: (i) mouse liver-derived hepatoblasts (ED 12.5-13), (ii) fetal (ED 17-18), (iii) perinatal (1 wk-old), (iv) adult (2 m-old) hepatocytes and cholangiocytes, and (v) adult mouse liver stem cell-derived progenitors. We have also developed an efficient in vitro differentiation protocol for human embryonic stem (ES) and induced pluripotent stem (iPS) cells that we will use for transformation. The tumors will be subjected to histological evaluation and integrated genomic analyses (integrating gene expression signatures with CNV and epigenetic analyses). We will also address whether the differentiation stage at transformation determines the properties and frequency of CSC.